no code implementations • 11 Jan 2024 • Antoine Dedieu, Wolfgang Lehrach, Guangyao Zhou, Dileep George, Miguel Lázaro-Gredilla
Despite their stellar performance on a wide range of tasks, including in-context tasks only revealed during inference, vanilla transformers and variants trained for next-token predictions (a) do not learn an explicit world model of their environment which can be flexibly queried and (b) cannot be used for planning or navigation.
no code implementations • 10 May 2023 • Dániel L Barabási, Ginestra Bianconi, Ed Bullmore, Mark Burgess, SueYeon Chung, Tina Eliassi-Rad, Dileep George, István A. Kovács, Hernán Makse, Christos Papadimitriou, Thomas E. Nichols, Olaf Sporns, Kim Stachenfeld, Zoltán Toroczkai, Emma K. Towlson, Anthony M Zador, Hongkui Zeng, Albert-László Barabási, Amy Bernard, György Buzsáki
We explore the challenges and opportunities in integrating multiple data streams for understanding the neural transitions from development to healthy function to disease, and discuss the potential for collaboration between network science and neuroscience communities.
no code implementations • 13 Mar 2023 • Miguel Lazaro-Gredilla, Ishan Deshpande, Sivaramakrishnan Swaminathan, Meet Dave, Dileep George
We consider the problem of recovering a latent graph where the observations at each node are \emph{aliased}, and transitions are stochastic.
no code implementations • 14 Feb 2023 • J. Swaroop Guntupalli, Rajkumar Vasudeva Raju, Shrinu Kushagra, Carter Wendelken, Danny Sawyer, Ishan Deshpande, Guangyao Zhou, Miguel Lázaro-Gredilla, Dileep George
Graph schemas can be learned in far fewer episodes than previous baselines, and can model and plan in a few steps in novel variations of these tasks.
1 code implementation • ICCV 2023 • Guangyao Zhou, Nishad Gothoskar, Lirui Wang, Joshua B. Tenenbaum, Dan Gutfreund, Miguel Lázaro-Gredilla, Dileep George, Vikash K. Mansinghka
In this paper, we introduce probabilistic modeling to the inverse graphics framework to quantify uncertainty and achieve robustness in 6D pose estimation tasks.
Ranked #1 on on YCB-Video
no code implementations • 31 Jan 2023 • Antoine Dedieu, Guangyao Zhou, Dileep George, Miguel Lazaro-Gredilla
We evaluate both approaches on several benchmarks where VI is the state-of-the-art and show that our method (a) achieves better test performance than Ji et al. (2020) for learning noisy-OR BNs with hierarchical latent structures on large sparse real datasets; (b) recovers a higher number of ground truth parameters than Buhai et al. (2020) from cluttered synthetic scenes; and (c) solves the 2D blind deconvolution problem from Lazaro-Gredilla et al. (2021) and variant - including binary matrix factorization - while VI catastrophically fails and is up to two orders of magnitude slower.
1 code implementation • 24 Jan 2023 • Ken Kansky, Skanda Vaidyanath, Scott Swingle, Xinghua Lou, Miguel Lazaro-Gredilla, Dileep George
We provide a benchmark of more than 200 PushWorld puzzles in PDDL and in an OpenAI Gym environment.
no code implementations • 3 Dec 2022 • Rajkumar Vasudeva Raju, J. Swaroop Guntupalli, Guangyao Zhou, Miguel Lázaro-Gredilla, Dileep George
Fascinating and puzzling phenomena, such as landmark vector cells, splitter cells, and event-specific representations to name a few, are regularly discovered in the hippocampus.
2 code implementations • 8 Feb 2022 • Guangyao Zhou, Antoine Dedieu, Nishanth Kumar, Wolfgang Lehrach, Miguel Lázaro-Gredilla, Shrinu Kushagra, Dileep George
PGMax is an open-source Python package for (a) easily specifying discrete Probabilistic Graphical Models (PGMs) as factor graphs; and (b) automatically running efficient and scalable loopy belief propagation (LBP) in JAX.
1 code implementation • 6 Dec 2021 • Guangyao Zhou, Wolfgang Lehrach, Antoine Dedieu, Miguel Lázaro-Gredilla, Dileep George
To demonstrate MAM's capabilities to capture CSIs at scale, we apply MAMs to capture an important type of CSI that is present in a symbolic approach to recurrent computations in perceptual grouping.
1 code implementation • NeurIPS 2021 • Miguel Lazaro-Gredilla, Antoine Dedieu, Dileep George
Perturb-and-MAP offers an elegant approach to approximately sample from a energy-based model (EBM) by computing the maximum-a-posteriori (MAP) configuration of a perturbed version of the model.
1 code implementation • 3 Dec 2020 • Antoine Dedieu, Miguel Lázaro-Gredilla, Dileep George
We consider the problem of learning the underlying graph of a sparse Ising model with $p$ nodes from $n$ i. i. d.
no code implementations • 11 Jun 2020 • Nishad Gothoskar, Miguel Lázaro-Gredilla, Dileep George
For an intelligent agent to flexibly and efficiently operate in complex environments, they must be able to reason at multiple levels of temporal, spatial, and conceptual abstraction.
1 code implementation • 11 Jun 2020 • Miguel Lázaro-Gredilla, Wolfgang Lehrach, Nishad Gothoskar, Guangyao Zhou, Antoine Dedieu, Dileep George
Here we introduce query training (QT), a mechanism to learn a PGM that is optimized for the approximate inference algorithm that will be paired with it.
no code implementations • 10 Mar 2020 • Nishad Gothoskar, Miguel Lázaro-Gredilla, Abhishek Agarwal, Yasemin Bekiroglu, Dileep George
Our method can handle noise in the observed state and noise in the controllers that we interact with.
no code implementations • 10 Feb 2020 • Daniel P. Sawyer, Miguel Lázaro-Gredilla, Dileep George
The ability of humans to quickly identify general concepts from a handful of images has proven difficult to emulate with robots.
no code implementations • pproximateinference AABI Symposium 2019 • Miguel Lazaro-Gredilla, Wolfgang Lehrach, Dileep George
We show that our approach generalizes to unseen probabilistic queries on also unseen test data, providing fast and flexible inference.
no code implementations • 4 Sep 2019 • Dileep George
This paper is the preprint of an invited commentary on Lake et al's Behavioral and Brain Sciences article titled "Building machines that learn and think like people".
no code implementations • 1 May 2019 • Antoine Dedieu, Nishad Gothoskar, Scott Swingle, Wolfgang Lehrach, Miguel Lázaro-Gredilla, Dileep George
We show that by constraining HMMs with a simple sparsity structure inspired by biology, we can make it learn variable order sequences efficiently.
no code implementations • 6 Dec 2018 • Miguel Lázaro-Gredilla, Dianhuan Lin, J. Swaroop Guntupalli, Dileep George
Humans can infer concepts from image pairs and apply those in the physical world in a completely different setting, enabling tasks like IKEA assembly from diagrams.
no code implementations • 3 Aug 2018 • Dileep George, Alexander Lavin, J. Swaroop Guntupalli, David Mely, Nick Hay, Miguel Lazaro-Gredilla
Understanding the information processing roles of cortical circuits is an outstanding problem in neuroscience and artificial intelligence.
2 code implementations • ICML 2017 • Ken Kansky, Tom Silver, David A. Mély, Mohamed Eldawy, Miguel Lázaro-Gredilla, Xinghua Lou, Nimrod Dorfman, Szymon Sidor, Scott Phoenix, Dileep George
The recent adaptation of deep neural network-based methods to reinforcement learning and planning domains has yielded remarkable progress on individual tasks.
no code implementations • CVPR 2017 • Austin Stone, Huayan Wang, Michael Stark, Yi Liu, D. Scott Phoenix, Dileep George
Convolutional neural networks (CNNs) have shown great success in computer vision, approaching human-level performance when trained for specific tasks via application-specific loss functions.
no code implementations • NeurIPS 2016 • Xinghua Lou, Ken Kansky, Wolfgang Lehrach, CC Laan, Bhaskara Marthi, D. Scott Phoenix, Dileep George
We demonstrate that a generative model for object shapes can achieve state of the art results on challenging scene text recognition tasks, and with orders of magnitude fewer training images than required for competing discriminative methods.
no code implementations • 8 Nov 2016 • Huayan Wang, Anna Chen, Yi Liu, Dileep George, D. Scott Phoenix
Neural networks have shown to be a practical way of building a very complex mapping between a pre-specified input space and output space.
no code implementations • 7 Nov 2016 • Miguel Lázaro-Gredilla, Yi Liu, D. Scott Phoenix, Dileep George
We introduce the hierarchical compositional network (HCN), a directed generative model able to discover and disentangle, without supervision, the building blocks of a set of binary images.